Copyright © 2026 Saccadic LLC · Patent pending.

Browse chapters
  1. Release NotesChanges by version
  2. Chapter 1What IRIS Is and Is Not
  3. Chapter 2The IRIS Data Model
  4. Chapter 3How IRIS Measures
  5. Chapter 4Safe Interpretation and the First 15 Minutes
  6. Chapter 5Dashboard Overview
  7. Chapter 6Installation
  8. Chapter 7Configuration
  9. Chapter 8Build the Model Spec and Probe Set
  10. Chapter 9Minimal Tracker Integration
  11. Chapter 10Launch the Bench and Verify Capture
  12. Chapter 11Runtime Control, Signals, Metrics, and Logs
  13. Chapter 12Integration Patterns by Model Family
  14. Chapter 13Distributed Training
  15. Chapter 14Artifacts, Reports, Export, and Security
  16. Chapter 15Troubleshooting
  17. Chapter 16Whole Model and Layer Structure
  18. Chapter 17Domain Training and Affinity
  19. Chapter 18Attention-Head Diagnostics
  20. Chapter 19MoE Expert and Router Diagnostics
  21. Chapter 20Facets and Emergent Pathways
  22. Chapter 21Gradient Residuals, Stability, Curvature, and Update Efficiency
  23. Chapter 22Adaptive Cadence and Event Detection
  24. Chapter 23CLI Reference
  25. Chapter 24Configuration-Key Reference
  26. Chapter 25API Reference
  27. Chapter 26File and Schema Reference
  28. Chapter 27Glossary and Concept Index
  29. Chapter 28The Flow Map
  30. Appendix ADiagnostic Availability Matrix
  31. Appendix BLLM Prompt Library
On this page

Chapter 6 — Installation

Use the instructions supplied with your distribution

Get IRIS™ from the Saccadic download dashboard and follow its Install IRIS instructions. The distribution's Installation Guide covers prerequisites, the package installation, account/run-key setup, and verification. Do not substitute an unrelated package with a similar name.

IRIS installs as one unit. The tracker, Bench, and optional Butler relay are parts of the same distribution, not separately selected installations.

Part Job
Tracker Runs inside the training process and writes diagnostic frames.
Bench Opens the run directory and presents the captured measurements.
Butler Optionally relays scalars to an experiment tracker you already use.

Install into the training environment

The interpreter running training must be the interpreter into which IRIS was installed. Activating a different environment can leave iris absent from the shell or produce No module named 'iris' in the training script. Installing on the machine alone is not enough.

Revision 2 lists these interpreter targets: Python 3.12, 3.13, or 3.14 on Apple silicon; Python 3.12 on Intel Mac and Linux x86_64. Follow the requirements shipped with your particular distribution if they differ.

Verify before a long job

iris --help
iris smoke --dryrun
iris --license

The dryrun smoke test checks installation wiring and licensing reachability without consuming credits or recording science. The full iris smoke test performs an end-to-end verification and presents its cost disclosure; read that disclosure before proceeding. iris --license checks readiness without starting a Bench or placing a credit hold.

Read startup messages before assuming capture is working. If the install reports that no bundle can be written, repair the named problem and rerun; successful model training does not prove that diagnostic frames were saved. See Troubleshooting.